- Object storage can help telcos manage massive amounts of unstructured data
- The ability to access data in a single location at the storage layer can help fuel AI pipelines
- Adopting object storage is also helping telcos simplify their software stacks
It’s one thing to run an AI proof of concept. It’s another thing entirely to run AI workloads in production at telco scale. Why? In part, because good AI feeds on good data. And finding a way to make the reams of unstructured data that telcos have available to AI can be tricky. That’s where object storage comes in.
Somewhere in the realm of 80% to 90% of data generated by organizations is unstructured, according to IDC and Gartner. Files like old emails, web pages, PDFs, purchase orders, invoices, training manuals, repair guides and even sensor readings – that’s all unstructured data and exactly the kind of information that AI could really use.
Object storage – like Amazon’s S3 and similar offerings from IBM, Wasabi, MinIO and a range of others – provides a way to manage and access these massive volumes of data via on-prem servers or the cloud. Think of it as the virtual equivalent to sorting all the stuff in your garage into vaguely labeled bins.
“It’s a surprisingly good fit for all the reasons you would think – data volume and sprawl, unstructured data, natural feed for AI data pipelines – these all speak to what a telco needs to consider in this AI era,” Moor Insights and Strategy VP and Principal Analyst Matt Kimball told Fierce.
Indeed, Backblaze CEO Gleb Budman said object storage is “increasingly the foundation of the data lake—and the simplest way to handle exploding data volumes.” Notably, it provides a single location to consolidate everything from logs and video to network data, and it can support both internal and external use cases.
“Simply put: no modern data lake, no AI at telco scale,” Budman said.
Operators wise up
Telcos seem to be wising up to this reality and are turning to object storage to innovate and modernize their software stacks.
Public records indicate that T-Mobile is using Amazon’s S3 storage (among other AWS services) as well as tools from Pure Storage’s Portworx. Additionally, MinIO co-CEO Garima Kapoor told Fierce the company counts both Verizon and Rakuten as customers, with the former using it for their private cloud backbone, messaging and voicemail services and the latter using MinIO as the foundation for its simplified OSS/BSS stack.
“From telco perspective specifically, I think they’re looking for things to become more AI ready than where they are right now,” Kapoor said.
Of Rakuten specifically, she added that their legacy file systems were not able to scale to the level that the telco wanted, which was what drove Rakuten to work with MinIO on rebuilding its stack to ensure it was software-defined and flexible.
As more telcos simplify their stack, “they can now let go of middleware and start to bring applications directly on top of object storage, much like how AWS has also built their stack,” she continued. That means letting go of middleware databases and query engines where they’re not needed and enabling AI agents to directly “start consuming this data through the Iceberg table format.”
Why it matters
Futuriom Senior Analyst Mary Jander noted that the ability to store telco data in a single location with unified access to object and table metadata at the storage layer “can significantly improve the efficiency and manageability of data in AI pipelines.” Kimball added such a move also offers opportunities to reduce operational complexity and costs and improve performance.
Speaking to Kapoor’s point about removing middleware and running AI applications directly on the object storage, Kimball said, “There is a time and place for indexing, schema enforcement, table compacting, etc. But all of these can take a toll on AI performance.”
Both Jander and Kimball added the ability to deploy object storage on-premises (as can be done with MinIO) also addresses telco concerns around privacy, security and data sovereignty.
Asked just how widespread telco interest in and adoption of object storage is, Kapoor said it’s still a work in progress.
“It’s a journey. A lot of things are happening. They are not there yet, but in terms of getting the foundation ready from an infrastructure standpoint, that’s how they are thinking of how to be AI ready,” she concluded.